If the JVM analysis didn't indicate a memory issue, finally the application is the dominating consumer of the CPU. This means that the application code itself is responsible for the bottleneck. Especially applications that run sophisticated algorithms excessively utilize the CPU.

Application-related profiling will lead to conclusions where in the application the issue originates and how the issue might be resolved. This means that the application either contains suboptimal code or reached the possible limit with the given resources, and ultimately needs to be scaled horizontally or vertically.